Peptidylarginine Deiminase 2-Catalyzed Histone H3 Arginine 26 Citrullination Facilitates Estrogen Receptor Α Target Gene Activation

Peptidylarginine deiminase 2-catalyzed histone H3 arginine 26 citrullination facilitates estrogen receptor α target gene activation

Xuesen Zhanga, Michael Boltb, Michael J. Guertinc, Wei Chend, Sheng Zhangd, Brian D. Cherringtona,1, Daniel J. Sladee, Christina J. Dreytone, Venkataraman Subramaniane, Kevin L. Bickere, Paul R. Thompsone, Michael A. Mancinib, John T. Lisc, and Scott A. Coonroda,2

aBaker Institute for Animal Health, College of Veterinary Medicine, cDepartment of Molecular Biology and Genetics, and dProteomics and Mass Spectrometry Core Facility, Cornell University, Ithaca, NY 14853; bDepartment of Molecular and Cellular Biology, Baylor College of Medicine, Houston, TX 77030; and eDepartment of Chemistry, The Scripps Research Institute, Jupiter, FL 33458

Edited by Jerry L. Workman, Stowers Institute for Medical Research, Kansas City, MO, and accepted by the Editorial Board July 2, 2012 (received for review February 24, 2012) Cofactors for estrogen receptor α (ERα) can modulate gene activity Results and Discussion by posttranslationally modifying histone tails at target promoters. Estrogen Induces H3R26 Citrullination in Cellulo and in Vivo. To begin Here, we found that stimulation of ERα-positive cells with 17β-estra- testing for associations between histone citrullination and E2 sig- diol (E2) promotes global citrullination of histone H3 arginine 26 naling, we first investigated whether estrogen stimulation globally (H3R26) on chromatin. Additionally, we found that the H3 citrulline induced citrullination of specific histone arginine residues in 26 (H3Cit26) modification colocalizes with ERα at decondensed chro- MCF-7 breast cancer cells by using confocal immunofluorescence matin loci surrounding the estrogen-response elements of target with three different site-specific anticitrullinated histone anti- promoters. Surprisingly, we also found that citrullination of H3R26 bodies: anti-H3Cit2/8/17, anti-H3Cit26, and anti-H4Cit3. Results is catalyzed by peptidylarginine deiminase (PAD) 2 and not by PAD4 showed that 45 min of E2 treatment induced a pronounced in- (which citrullinates H4R3). Further, we showed that PAD2 interacts crease of H3Cit26 in the nuclei of cells (Fig. S1), whereas staining with ERα after E2 stimulation and that inhibition of either PAD2 or with the two other anticitrullinated histone antibodies was not ERα strongly suppresses E2-induced H3R26 citrullination and ERα visibly affected (Fig. S2 A and B). Interestingly, after E2 stimu- recruitment at target gene promoters. Collectively, our data suggest lation, we also did not observe global differences in three other that E2 stimulation induces the recruitment of PAD2 to target pro- histone modifications known to be altered at specific promoters by moters by ERα, whereby PAD2 then citrullinates H3R26, which leads E2: H4K5acteyl, H3K9dimethyl, and H3K27trimethyl (Fig. S2 C– to local chromatin decondensation and transcriptional activation. E). Further, the increase in H3Cit26 levels was observed as early as 5 min after E2 stimulation and appeared to peak at 45 min (Fig. ancers of the female reproductive system are serious human S3). This time frame is consistent with recent reports on the dy- Chealth problems, and estrogen plays a critical role in the namics of estrogen signaling whereby changes in the MCF-7 cell initiation and progression of these diseases (1). Despite decades transcriptome were observed by GRO-seq analysis within 10 min of research into mechanisms of 17β-estradiol (E2)-responsive of E2 stimulation (14). gene transcription, our understanding of this process is far from Given these exciting results, and the relatively uncharacterized complete (2). It is generally believed that, upon E2 binding, the nature of the anti-H3Cit26 antibody, we next further validated the nuclear hormone receptor estrogen receptor α (hereafter called specificity of this antibody by treating MCF-7 cell histones with ER) undergoes major structural reorganization, associates with recombinant human PAD2 and then immunoblotting the re- estrogen-response elements (ERE) within target gene pro- solved proteins with the H3Cit26 antibody (Note: Our rationale moters, and recruits a range of coactivators including histone for using PAD2 as opposed to PAD4 is described below). Results modification enzymes (3–6). After deposition, the resulting his- (Fig. S4A) show that this antibody was reactive with an appro- tone modifications can then modulate target gene activity by priately sized band from the PAD2-citrullinated histones but was affecting local chromatin structure and regulating the accessi- not reactive with untreated histones. Further, we also found that bility of chromatin to transcription factors (2, 5, 7–9). preincubation of the anti-H3Cit26 antibody with the cognate Peptidylarginine deiminase (PAD) enzymes convert arginine citrullinated peptide nearly completely blocked detection of the fi and methylarginine residues to citrulline via a hydrolytic process H3Cit26 modi cation. The protein band corresponding to the termed citrullination or deimination (10, 11). We and others mass of the citrullinated histone was then excised from the gel and CELL BIOLOGY have shown that one such PAD, PAD4, appears to play a re- evaluated by mass spectroscopic (MS) analysis. Results show that pressive role in regulating the expression of the canonical ER citrulline was readily detected on histone H3 peptides (24-AAR target gene, TFF1, via citrullination of histone H4 methylarginine 3, thus suggesting that PADs potentially function as ER cofactors (12, 13). Given that these previous studies were limited Author contributions: X.Z. and S.A.C. designed research; X.Z., M.B., B.D.C., D.J.S., C.J.D., V.S., K.L.B., and P.R.T. performed research; X.Z., M.J.G., W.C., S.Z., M.A.M., and J.T.L. to a single ER target promoter, we chose to take a more com- analyzed data; and X.Z., M.J.G., J.T.L., and S.A.C. wrote the paper. prehesive approach to test whether PAD-mediated histone tail The authors declare no conflict of interest. citrullination may be more fundamental to ER target gene reg- This article is a PNAS Direct Submission. J.L.W. is a guest editor invited by the Editorial ulation than previously realized. In this study, we show that cit- Board. rullination of histone H3R26 at ER targets is closely associated Data deposition: The data reported in this paper have been deposited in the Gene Ex- with gene transcription and that citrullination at this residue is pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE32599). catalyzed by PAD2, as opposed to PAD4. Additionally, we show 1Present address: Department of Zoology and Physiology, University of Wyoming, Lara- that PAD2 interacts with ER and that PAD2-mediated citrulli- mie, WY 82071. nation of H3R26 likely facilitates transcriptional activation by 2To whom correspondence should be addressed. E-mail: [email protected]. creating an open, permissive, chromatin architecture around the This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. EREs of E2-induced genes. 1073/pnas.1203280109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1203280109 PNAS | August 14, 2012 | vol. 109 | no. 33 | 13331–13336 Downloaded by guest on September 23, 2021 [Cit]K[acetyl]SAPATGGVK-36) from the PAD2-treated sample either before or after E2 treatment. However, in HeLa cells that (Fig. S4B). We note here that the H3Cit26 modification only stably overexpress recombinant ER, E2 stimulation resulted in occurred on H3 peptides that contained an acetyl modification at a robust citrullination of H3R26 (Fig. 1E). Thus, our data strongly lysine-27, thus raising the possibility that a functional “cross talk” suggest citrullination of H3R26 depends on E2-ER signaling. exists between the H3Cit26 and H3K27acetyl modifications. To We next investigated the dynamics of estrogen-ER–mediated confirm that the anti-H3Cit26 antibody was not recognizing other H3R26 citrullination by using another HeLa cell line that con- citrullinated residues on chromatin, we also preblocked the anti- tains multiple genomically integrated copies of the estrogen-re- H3Cit26 antibody by using unmodified H3 (19–38), H3Cit26, or sponsive prolactin (PRL) enhancer/promoter reporter array (Fig. H3Cit2/8/17 peptides and then performed chromatin immuno- 1F) and a stably expressed ER-GFP-fusion construct (GFP-ER) precipitation (ChIP) analysis on the TFF1 ERE promoter region. (15). This line allows for the imaging of GFP-ER binding to the Results (Fig. S4C) showed that only the H3Cit26 peptide abol- PRL array and subsequent large-scale changes in chromatin ished the detection of H3Cit26 at the TFF1 ERE. Taken together, structure. Results from previous studies found that E2 stimula- these results strongly suggest that the H3Cit26 antibody is spe- tion induces an ER-dependent decondensation of the PRL array, cifically reactive with the H3Cit26 modification. thus reflecting a chromatin state similar to that found at endog- To investigate whether our findings may have physiological enous ER target promoters during transactivation. Interestingly, relevance in vivo, we next treated ovariectomized mice with ex- previous studies demonstrated that levels of specific histone ogenous E2 pellets and then carried out immunohistochemical acetylation and methylation marks are not greatly altered at the analysis of uterine tissue (a major E2 target) by using the anti- PRL array after E2 stimulation (16). In our study, before E2 H3Cit26 antibody. As predicted, relative to vehicle-treated ani- stimulation, little citrullination of H3R26 was observed. However, mals, we observed a strong increase in H3Cit26 staining in the after E2 treatment, a large H3Cit26 foci was found to strongly nucleus of uterine epithelial cells after E2 treatment (Fig. 1A). colocalize with GFP-ER at the decondensed PRL array (Fig. 1F Therefore, both our in vitro and in vivo observations strongly and Fig. S5). Given our recent finding that PAD-dependent his- suggest that citrullination of histone H3R26 is specifically and tone hypercitrullination results in chromatin decondensation globally induced by estrogen signaling. (17), we predicted that the targeting of ER to the PRL array by estrogen likely resulted in ER-mediated recruitment of PADs to Estrogen-Induced H3R26 Citrullination Is Associated with ER Activity. the promoter and subsequent PAD-mediated citrullination of Given that the effects of E2 on gene regulation are primarily H3R26, thus resulting in decondensation of the PRL array. mediated by ER, we next investigated the relationship between H3Cit26 and ER in cells. After E2 stimulation, immunofluores- ERE Motifs Overlap H3Cit26 Sites at E2-Induced Gene Promoters. cence analysis showed a strong colocalization between H3Cit26 Given the observed global associations between histone citrulli- and ER in MCF-7 cells (Fig. 1B). In ER-negative MDA-MB231 nation and estrogen signaling, we next began to investigate the breast cancer cells, however, citrullination of H3R26 was not extent to which H3R26 citrullination correlates with ER binding observed, either before or after E2 stimulation (Fig. 1C). We at proximal promoter target regions by conducting ChIP/ further confirmed the requirement of ER for H3R26 citrullina- promoter analysis in MCF-7 cells using a tiling array that covers tion by using HeLa cells, which are also ER negative (Fig. 1D). As 2.2 kb upstream and 500 bp downstream, relative to the transcrip- expected, H3R26 citrullination was not observed in these cells tion start sites (TSS) (18). Comparison of citrullination at H3R26

Fig. 1. Estrogen-induced H3R26 citrullination is associated with ER. (A) IHC analysis of H3Cit26 staining on uterine sections collected from ovariectomized mice implanted with pellets containing either vehicle control or E2 pellets. (Scale bars: 20 μm.) (B–E) Confocal microscopic analyses showing colocalization of H3Cit26 and ER in nuclei of ER-positive MCF-7 cells (B). This colocalization is not observed in either ER-negative MDA-MB231 (C) or HeLa (D) cells upon E2 stimulation. Stable overexpression of ER in HeLa cells (E) restores E2-induced citrullination of H3R26. Merged images highlight H3Cit26 and ER colocalization. (Scale bar: 10 μm.) (F) Colocalization of H3Cit26 and GFP-ER at a decondensed PRL-array in HeLa cells stably overexpressing GFP-ER. Arrows indicate GFP-ER, and arrowheads indicate H3Cit26 foci in the decondensed array after E2 stimulation. (Scale bar: 15 μm.) Inset highlights a single PRL-array locus. (Magniﬁcation: 10×.) Yellow arrowhead represents the coincident anti-H3Cit26 (red) and GFP- ER (green) staining. Schema in Lower shows the essential elements of the PRL-based array (16). Xn indicates 52 repeats of the elements.

13332 | www.pnas.org/cgi/doi/10.1073/pnas.1203280109 Zhang et al. Downloaded by guest on September 23, 2021 before and after E2 stimulation revealed 208 promoters that of ER binding sites are within 5 kb upstream of the TSS (22). contained significantly higher levels of the H3Cit26 modification Given that the proximal promoter arrays used in our study only after E2 treatment (induced), 110 promoters with lower H3Cit26 covered 2.2 kb upstream of the TSS, it seems likely that many ER levels after E2 treatment (reduced), and 55 promoters with no target sites outside of this region are also citrullinated at H3R26 changes in H3Cit26 levels (constitutive) (Fig. 2 A and B). To after E2 treatment. further characterize the putative targets of H3Cit26-regulated gene transcription, we used de novo motif discovery to search for H3R26 Citrullination Likely Promotes an Open Chromatin Architecture overrepresented motifs within 100 bp of the center of the H3Cit26 at the EREs of E2-Induced Gene Promoters. A growing body of evi- peaks (19). This unbiased analysis revealed the presence of dence suggests that specific histone tail modifications function, a previously published, well-characterized ERE motif (20) (Fig. in part, by “loosening” the chromatin architecture at gene pro- 2C). The heatmap in Fig. 2D demonstrates that the ERE motif moters, thus enhancing accessibility for binding factors, such as displayed a similar distribution pattern as that of H3Cit26 in the basal and specific transcription factors and RNA polymerases (2, induced group (top left to lower right diagonal), and this pattern 5, 7–9). Micrococcal nuclease (MNase) cleaves linker and nu- was not observed in the noninduced groups (reduced and con- cleosome-free DNA, and is, therefore, often used to evaluate the stitutive). We identified 59 targets (28.4%) as having an ERE in state of chromatin compaction (23, 24). To test the hypothesis the H3Cit26-induced group, whereas only 8 (4.8%) genes in the that H3R26 citrullination might act to maintain an open chro- noninduced groups contained the ERE motif, thus indicating matin architecture around ERE regions at the promoters of a a 5.9-fold ERE enrichment in the H3Cit26-induced group subset of E2-induced genes, we next examined the accessibility of − (Fisher’s exact test, P = 7.5 × 10 10) (Fig. 2E). Collectively, these ERE-containing H3Cit26-bound promoters to digestion by data support the hypothesis that E2-induced citrullination at MNase as described (23, 24). We tracked five ER targets from H3R26 plays a role modulating the expression of a subset of ER the induced group that contain well-characterized EREs in their target genes. Given that ER can bind to chromatin through promoters, TFF1 (25), KRT13 (26), GREB1 (27), WISP2 (28), bridging interactions with other transcription factors, such as AP- and CYP1B1 (29, 30) and also two control genes, SCN1A and 1, for example, and that these ER binding sites often do not VAT1L, which did not show an increase or decrease in citrulli- contain underlying EREs (21), it seems likely that the H3Cit26 nation after E2 treatment and do not appear to contain an ERE modification is also enriched at other non-ERE–containing target near the citrullinated site (i.e., the noninduced group). As sites. Additionally, a recent ChIP-seq report found that only ∼7% expected, E2 treatment caused a significant decrease in MNase

Fig. 2. Genome-wide location analysis of transcriptional targets for H3R26 citrullination in MCF-7 cells reveals ERE motifs overlapping the H3Cit26 CELL BIOLOGY sites at E2-induced gene promtoers. (A) Heat maps of anti-H3Cit26 ChIP-chip data showing the signiﬁ- cant H3Cit26 peaks across target promoters from −2200 to +500 bp relative to the TSS before (EtOH Control) and after E2 treatment. The genes in each group (y axis) are ordered from those with the 5′- most H3Cit26 peak to those with the 3′-most peak. (B) Averaging of peak centered ChIP-chip data across the groups in A illustrates the distinct pat- terns of H3R26 citrullination at target promoters in response to E2. (C) De novo determination of DNA sequence motif associated with H3Cit26 at the E2- induced targets reveals a consensus ERE motif. C Lower shows a published ERE motif (20). (D) Heat maps of signiﬁcant EREs that overlap with the H3Cit26 peak. EREs are ordered according to H3Cit26 heat maps. (E) Percentage and fold enrichment of EREs in the E2-induced and noninduced H3Cit26 datasets.

Zhang et al. PNAS | August 14, 2012 | vol. 109 | no. 33 | 13333 Downloaded by guest on September 23, 2021 protection at the ERE region on the promoters of the E2-in- at EREs. As shown in Fig. 4 A and B, although E2 is able to induce duced genes relative to the EtOH control (Fig. 3 A and B and both H3R26 citrullination and ER recruitment at TFF1 and Fig. S6 A–C), and this effect was not observed at the noninduced KRT13, inhibiting ER blocked not only ER recruitment, but also control promoters (Fig. 3 A and B and Fig. S6D). This obser- strongly suppressed H3R26 citrullination at this region. H3R26 vation suggests that, after E2 treatment, H3R26 citrullination at citrullination at the SCN1A promoter was not affected by any the ERE regions of E2-induced gene promoters results in a treatment. These data indicate that E2-induced recruitment of ER MNase hypersensitivity profile that is indicative of an open to a subset of responsive gene promoter EREs facilitates H3R26 chromatin architecture. Moreover, ChIP analysis (Fig. 3 C and citrullination at these sites. D) revealed that, after E2 stimulation, there were marked We then wondered whether decreased H3R26 citrullination increases in ER binding and H3R26 citrullination at the ERE could have a similar inhibitory effect on ER recruitment. To regions on TFF1 and KRT13 promoters. Although E2 weakly address this question, however, we first needed to identify which induced the recruitment of ER to SCN1A, increased citrullina- PAD was responsible for catalyzing H3R26 citrullination. Al- tion at H3R26 was not observed, thus further confirming our though most previous reports have documented a role for PAD4 promoter array analysis. Taken together, our data suggest that in gene regulation via histone citrullination (12, 13), we have re- E2-induced citrullination of H3R26 helps to establish an open cently found that PAD2 also appears to citrullinate histones in vivo and active chromatin environment at ER target promoters. (31). This finding raised the possibility that the observed citrullination at H3R26 might be mediated by PAD2 as opposed to Interdependent PAD2-Catalyzed H3R26 Citrullination and ER Recruitment PAD4. To test this hypothesis, we first evaluated the substrate at Induced Target Promoter EREs Facilitates Target Gene Transcription. preferences of these enzymes by determining the steady-state ki- We next pretreated MCF-7 cells with the pure ER antagonist, netic parameters for histones H3 and H4, as well as several pep- ICI182780, and then performed ChIP to test for potential E2-in- tides whose sequences are based on the N-termini of these proteins duced functional interplay between ER and H3R26 citrullination and encompass residues known to be citrullinated (Table S1).

Fig. 3. After estrogen stimulation, H3R26 cit- rullinaton facilitates an open chromatin architecture at the EREs of E2-induced gene promoters. (A) MNase protection assay after EtOH and E2 treatment in MCF-7 cells. qPCR was performed to tile through the proximal promoter ERE region with overlapping amplicons (∼100-bp PCR product average, with ∼20 bp overlap). Relative ratio of the amount of digested DNA to genomic control was used to determine the extent of MNase protection. Values from overlapping primer sets are averaged. Each point represents the mean + SEM, n = 3. (B) H3Cit26 ChIP-chip signals localize to the proximal promoter ERE regions of E2-induced targets. The shaded region indicates the range tested in A for MNase protection assay. (C and D) ChIP-qPCR analysis of H3Cit26 and ER at target gene promoters after EtOH or E2 treatment in MCF-7 cells. Error bars indicate SEM, n = 3.

13334 | www.pnas.org/cgi/doi/10.1073/pnas.1203280109 Zhang et al. Downloaded by guest on September 23, 2021 TFF1 and KRT13 after E2 stimulation (Fig. 4E). Similar effects, however, were not observed in the PAD4 depleted line (Fig. S7). Taken together, our in vitro and in cellulo findings support our hypothesis that PAD2 (as opposed to PAD4) plays a direct role in citrullinating H3R26 in cells. Importantly, we also found that PAD2 depletion markedly suppressed ER recruitment to EREs at the TFF1 and KRT13 promoters (Fig. 4F). This result indicates that PAD2-mediated citrullination of H3R26 is likely to be in- volved in recruitment of ER to the ERE. One interpretation of our cumulative findings is that PAD2- mediated H3R26 citrullination opens the local chromatin architecture, thus facilitating ER recruitment. Alternatively, however, it is also possible that ER recruitment may induce a chromatin remodeling event that, in turn, yields a chromatin template that is accessible to citrullination at H3R26. In light of the recent “assisted loading” hypothesis (32), it is possible that both models are functional and this dual activity produces a reinforcing loop that ultimately enhances ER binding that, in turn, recruits the requisite coactivator complexes that create the appropriate chromatin environment for binding by the general transcriptional machinery and RNA polymerase. The interdependent nature of E2-stimulated H3R26 citrullination and ER recruitment at induced gene promoters also suggests a potential association between PAD2 and ER. To test this hypothesis, we first performed coimmunoprecipitation analysis in MCF-7 cells and found that endogenous PAD2 does appear to interact with endogenous ER and that this interaction is enhanced after E2 treatment (Fig. S8A). Reciprocally, we also generated an MCF-7 cell line that stably overexpresses a Flag-tagged version of PAD2 and found that E2 stimulates the interaction between endogenous ER and Flag-PAD2 (Fig. S8 B and C). We note here that the Flag-tagged PAD2 approach was necessary because immunoprecipitation- quality PAD2 antibodies are not available. We then carried out ChIP analysis by using the anti-Flag antibody in this modified MCF-7 cell line and found that, after E2 stimulation, Flag-tagged PAD2 was specifically recruited to the same sites within the ERE region of TFF1 that are bound by ER and contain the H3Cit26 modification (Fig. S8D). The observations that E2-induced chromatin decondensation is closely related to transcriptional activity (33) and that the H3Cit26 modification was primarily observed at the ERE regions of E2-induced genes suggested that PAD2-mediated citrullination of H3R26 at ERE promoter Fig. 4. Interdependent H3R26 citrullination catalyzed by PAD2 and ER regions facilities ER-mediated gene transactivation. To test this targeting at induced target promoter EREs facilitates target gene tran- hypothesis, we investigated whether the transcription of genes, scription after stimulation of MCF-7 cells with E2. (A and B) ChIP-qPCR whose promoters were citrullinated after E2 treatment, was af- analysis of H3Cit26 and ER at target gene promoters showing that complete fected by either PAD2 depletion or inhibition. As expected, either inhibition of ER by ICI182780 (ICI) also inhibits H3R26 citrullination at EREs. knockdown of PAD2 (Fig. 4G) or pretreatment of cells with Cl- fi (C) Plots of the ratio of the catalytic ef ciencies (kcat/Km(PAD2) / kcat/Km(PAD4)) Amidine, a newly developed arginine-based PAD inhibitor (34) fi showing the substrate speci city of PAD2 and PAD4. (D) Western blot (Fig. 4H), dramatically dampened the ability of these ER targets showing shRNA-mediated depletion of PAD2 in MCF-7 cells. (E and F) ChIP- qPCR analysis of H3Cit26 and ER at target gene promoters showing that to be activated after E2 treatment. CELL BIOLOGY depletion of PAD2 not only decreases H3Cit26 but also suppresses ER re- Taken together, our data indicate that PAD2 plays an impor- cruitment to EREs at target gene promoters. (G and H) Relative mRNA ex- tant role in mediating the activation of ER target genes via cit- pression (RT-qPCR) analyses in EtOH or E2-stimulated MCF-7 cells with or rullination of histone H3R26. This activity likely cooperates with without PAD2 knockdown (G); with or without Cl-Amidine treatment (H). other activities such as cofactor binding, chromatin remodeling Expression data were normalized to GAPDH transcripts, and the graph factor association, and basal transcription factor/RNA poly- represents the relative fold enrichment over vehicle treated. Error bars merase recruitment in establishing an open, permissive, chro- = indicate SEM, n 3. matin architecture around the EREs of E2-induced genes, thus facilitating transcriptional activation. These findings help to im- Analysis of the ratio of catalytic efficiencies (k /K / k / prove our mechanistic understanding of how ER regulates gene cat m(PAD2) cat transcription via altering chromatin structure. Km(PAD4)) reveals that PAD2 citrullinates the H3 (22–30) peptide, which contains H3R26, at a sixfold higher rate than PAD4 Materials and Methods (Fig. 4C), indicating that the histone H3R26 residue is a valid Cell Culture. Cell culture, shRNA, and Flag-tagged PAD2 overexpression are PAD2 target in vitro. To test whether PAD2 might target H3R26 in cellulo D described in SI Materials and Methods. Before E2 treatment, cells were , we then generated PAD2- (Fig. 4 ) and PAD4- (18) cultured for 3 d in DMEM phenol red-free medium supplemented with 10% depleted MCF-7 cell lines and performed ChIP analysis on the (vol/vol) charcoal-dextran-treated calf serum. ICI182780 was used at 10 μM TFF1 and KRT13 promoters. Results from the PAD2-depleted for 18 h before the addition of E2. Cl-Amidine was used at 200 μM for 40 h line showed a marked suppression of H3R26 citrullination at before E2 stimulation.

Zhang et al. PNAS | August 14, 2012 | vol. 109 | no. 33 | 13335 Downloaded by guest on September 23, 2021 Mouse Ovariectomy, E2 Treatment, and Immunohistochemistry (IHC). All pro- and subsequently used to make the sequence logo for ERE (Fig. 2C) using the cedures were conducted in accordance with the National Institutes of Health R package “seqLogo” contributed by Oliver Bembom (University of Cal- regulations and approved by the Cornell University animal use committee. ifornia, Berkeley, CA). For the heat map to visualize the ERE distribution, we See SI Materials and Methods for detailed protocols. used a published position-specific weight matrix (20) and searched for the matched that conform with a P value ≤0.00005. The promoters were sepa- Confocal Microscopy. Cells grown on slides were subjected to E2 treatment for rated into nonoverlapping 50-bp windows and, if an ERE motif was found 45 min. Confocal microscopy experiments were described (13). Antibodies within a window, the window was colored yellow. used are listed in SI Materials and Methods. MNase Protection Assay. Estrogen-starved MCF-7 cells were treated with High Content Analysis-Based Immunofluorescence Microscopy. Stable GFP-ERα: ethanol or 100 nM E2 for 45 min, and MNase protection assay was performed PRL-HeLa cells were E2 starved and then treated for 30 min with either as described (23, 24). MNase was from New England Biolabs (M0247). See fl fl 10 nM E2 or ethanol. Immuno uorescence experiment, uorescent micros- primers in SI Materials and Methods. copy image acquisition, and quantitation was performed as described (15). Nuclear masks were created by using the DAPI channel, and GFP-ER signal Gene-Specific Expression Analyses. Estrogen-starved cells were treated with was used to define the for PRL array mask. Average intensity measurements ethanol or 100 nM E2 for 6 h. RNA reverse transcription and quantitative real- were taken (>1,000 cells per treatment) for the nucleoplasmic and array time PCR were performed as described (18). Primers used are listed in SI H3Cit26 signals (in E2) and normalized to EtOH nucleoplasm H3Cit26 signal. Materials and Methods.

ChIP and ChIP-chip. ChIP experiments were performed as described (18). E2 Steady-State Kinetic Assays. Histone peptides were synthesized by using the was used at concentration of 100 nM for 45 min; antibodies and quantitative Fmoc approach and purified by reverse-phase HPLC. PADs 2 and 4 purifica- PCR (qPCR) primers are listed in SI Materials and Methods. ChIP for H3Cit26 coupled with hybridization to a human HG18 RefSeq promoter microarray tion, kinetic assays were performed as described in SI Materials and Methods. from Nimblegen and genomic data analyses were performed as described (18). Significant peaks and the “induced,”“reduced,” or “constitutive” Histone Extraction, PAD Assay, Mass Spectrometry, and Immunoprecipitation regions were defined as described in SI Materials and Methods. The TSS- Assay. See SI Materials and Methods for detailed protocols. anchored ChIP-chip heat maps were generated by using 600-bp windows with 150-bp steps and were visualized with Java Treeview (35). The data can ACKNOWLEDGMENTS. We thank W. L. Kraus for HeLa-ER cells and C. D. Allis fi be accessed through the NCBI/GEO website by using accession no. GSE32599. for the H3 unmodif ed peptide. This work was supported by a Department of Defense Era of Hope Scholar Award W871XWH-07-1-0372 (to S.A.C.), and National Institutes of Health Grants GM25232 (to J.T.L.) and GM079357 (to De Novo Motif Search. MEME was applied with all default parameters to P.R.T.). X.Z. was supported by a Postdoctoral Fellowship KG101303 from search for overrepresented motifs (19). A motif width between 6 and 20 bp Susan G. Komen for the Cure, and M.B. was supported by Training Fellow- was specified by using 200-bp windows centered on each of the 208 induced ship T15LM007093 from the Keck Center National Library of Medicine Train- H3Cit26 binding sites. Table S2 includes the matrix that was found by MEME ing Program in Biomedical Informatics of the Gulf Coast Consortia.

1. Cooke PS, Buchanan DL, Lubahn DB, Cunha GR (1998) Mechanism of estrogen action: 20. Grober OM, et al. (2011) Global analysis of estrogen receptor beta binding to breast Lessons from the estrogen receptor-alpha knockout mouse. Biol Reprod 59:470–475. cancer cell genome reveals an extensive interplay with estrogen receptor alpha for 2. Green KA, Carroll JS (2007) Oestrogen-receptor-mediated transcription and the in- target gene regulation. BMC Genomics 12:36. fluence of co-factors and chromatin state. Nat Rev Cancer 7:713–722. 21. Paech K, et al. (1997) Differential ligand activation of estrogen receptors ERalpha and 3. Brzozowski AM, et al. (1997) Molecular basis of agonism and antagonism in the ERbeta at AP1 sites. Science 277:1508–1510. oestrogen receptor. Nature 389:753–758. 22. Welboren WJ, et al. (2009) ChIP-Seq of ERalpha and RNA polymerase II defines genes 4. Mann M, Cortez V, Vadlamudi RK (2011) Epigenetics of estrogen receptor signaling: differentially responding to ligands. EMBO J 28:1418–1428. Role in hormonal cancer progression and therapy. Cancers (Basel) 3:1691–1707. 23. Krishnakumar R, Kraus WL (2010) PARP-1 regulates chromatin structure and tran- 5. Métivier R, et al. (2003) Estrogen receptor-alpha directs ordered, cyclical, and com- scription through a KDM5B-dependent pathway. Mol Cell 39:736–749. binatorial recruitment of cofactors on a natural target promoter. Cell 115:751–763. 24. Petesch SJ, Lis JT (2008) Rapid, transcription-independent loss of nucleosomes over – 6. Wiench M, Miranda TB, Hager GL (2011) Control of nuclear receptor function by local a large chromatin domain at Hsp70 loci. Cell 134:74 84. fi chromatin structure. FEBS J 278:2211–2230. 25. Shang Y, Hu X, DiRenzo J, Lazar MA, Brown M (2000) Cofactor dynamics and suf - – 7. Guertin MJ, Lis JT (2010) Chromatin landscape dictates HSF binding to target DNA ciency in estrogen receptor-regulated transcription. Cell 103:843 852. elements. PLoS Genet 6:e1001114. 26. Sheng S, Barnett DH, Katzenellenbogen BS (2008) Differential estradiol and selective 8. Li B, Carey M, Workman JL (2007) The role of chromatin during transcription. Cell 128: estrogen receptor modulator (SERM) regulation of Keratin 13 gene expression and its underlying mechanism in breast cancer cells. Mol Cell Endocrinol 296:1–9. 707–719. 27. Sun J, Nawaz Z, Slingerland JM (2007) Long-range activation of GREB1 by estrogen 9. Strahl BD, Allis CD (2000) The language of covalent histone modifications. Nature 403: receptor via three distal consensus estrogen-responsive elements in breast cancer 41–45. cells. Mol Endocrinol 21:2651–2662. 10. Klose RJ, Zhang Y (2007) Regulation of histone methylation by demethylimination 28. Fritah A, Redeuilh G, Sabbah M (2006) Molecular cloning and characterization of the and demethylation. Nat Rev Mol Cell Biol 8:307–318. human WISP-2/CCN5 gene promoter reveal its upregulation by oestrogens. J Endo- 11. Zhang Y (2004) Molecular biology: No exception to reversibility. Nature 431:637–639. crinol 191:613–624. 12. Cuthbert GL, et al. (2004) Histone deimination antagonizes arginine methylation. Cell 29. Han W, Pentecost BT, Pietropaolo RL, Fasco MJ, Spivack SD (2005) Estrogen receptor 118:545–553. alpha increases basal and cigarette smoke extract-induced expression of CYP1A1 and 13. Wang Y, et al. (2004) Human PAD4 regulates histone arginine methylation levels via CYP1B1, but not GSTP1, in normal human bronchial epithelial cells. Mol Carcinog 44: – demethylimination. Science 306:279 283. 202–211. 14. Hah N, et al. (2011) A rapid, extensive, and transient transcriptional response to es- 30. Tsuchiya Y, et al. (2004) Human CYP1B1 is regulated by estradiol via estrogen re- – trogen signaling in breast cancer cells. Cell 145:622 634. ceptor. Cancer Res 64:3119–3125. 15. Ashcroft FJ, Newberg JY, Jones ED, Mikic I, Mancini MA (2011) High content imaging- 31. Cherrington BD, Morency E, Struble AM, Coonrod SA, Wakshlag JJ (2010) Potential α fi based assay to classify estrogen receptor- ligands based on de ned mechanistic role for peptidylarginine deiminase 2 (PAD2) in citrullination of canine mammary – outcomes. Gene 477:42 52. epithelial cell histones. PLoS ONE 5:e11768. 16. Sharp ZD, et al. (2006) Estrogen-receptor-alpha exchange and chromatin dynamics 32. Voss TC, et al. (2011) Dynamic exchange at regulatory elements during chromatin are ligand- and domain-dependent. J Cell Sci 119:4101–4116. remodeling underlies assisted loading mechanism. Cell 146:544–554. 17. Wang Y, et al. (2009) Histone hypercitrullination mediates chromatin decondensation 33. Vic P, Garcia M, Humeau C, Rochefort H (1980) Early effect of estrogen on chromatin and neutrophil extracellular trap formation. J Cell Biol 184:205–213. ultrastructure in endometrial nuclei. Mol Cell Endocrinol 19:79–92. 18. Zhang X, et al. (2011) Genome-wide analysis reveals PADI4 cooperates with Elk-1 to 34. Luo Y, et al. (2006) Inhibitors and inactivators of protein arginine deiminase 4: activate c-Fos expression in breast cancer cells. PLoS Genet 7:e1002112. Functional and structural characterization. Biochemistry 45:11727–11736. 19. Bailey TL, et al. (2009) MEME SUITE: Tools for motif discovery and searching. Nucleic 35. Saldanha AJ (2004) Java Treeview—extensible visualization of microarray data. Acids Res 37(Web Server issue):W202-8. Bioinformatics 20:3246–3248.

13336 | www.pnas.org/cgi/doi/10.1073/pnas.1203280109 Zhang et al. Downloaded by guest on September 23, 2021